The Self-Fulfilling Prophecy in Close Relationships: Rejection

Journal of Personality and Social Psychology Copyright 1998 by the American Psychological Association, Inc. 1998, Vol. 75, No. 2, 545-560 0022-3514/98/$3,00

The Self-Fulfilling Prophecy in Close Relationships: Rejection Sensitivity and Rejection by Romantic Partners

Geraldine Downey, Antonio L. Freitas, Benjamin Michaelis, and Hala Khouri Columbia University

The authors hypothesized a self-fulfilling prophecy wherein rejection expectancies lead people to behave in ways that elicit rejection from their dating partners. The hypothesis was tested in 2 studies of conflict in couples: (a) a longitudinal field study where couples provided daily-diary reports and (b) a lab study involving behavioral observations. Results from the field study showed that high rejection-sensitive (HRS) people's relationships were more likely to break up than those of low rejection-sensitive (LRS) people. Conflict processes that contribute to relationship erosion were revealed for HRS women but not for HRS men. Following naturally occurring relationship conflicts, HRS women's partners were more rejecting than were LRS women's partners. The lab study showed that HRS women's negative behavior during conflictual discussions helped explain their partners' more rejecting postconflict responses.

People's beliefs about their significant others are assumed to influence the course of their relationships in important ways (for reviews, see Baldwin, 1992; Bradbury & Fincham, 1990; Hazan & Shaver, 1994; Reis & Patrick, 1996). One class of beliefs--expectations concerning acceptance and re ject ion-- has long been deemed especially vital to people's relationship functioning (e.g., Bowlby, 1969, 1973, 1980; Erikson, 1950; Homey, 1937; Sullivan, 1953). Bowlby ( 1969, 1973, 1980), for example, theorized that people's internal working models of relationships, incorporating expectations of rejection and acceptance, shape their relationships. In an example of what Merton (1948) termed the self-fulfilling prophecy, Sroufe (1990) suggested that rejection expectations can lead people to behave in ways that elicit rejection from others. In this article we examine whether and how this proposed self-fulfilling prophecy operates in the romantic relationships of people high in rejection sensitivity (RS).

Drawing selectively on attachment and social-cognitive ap- proaches to close relationships, we have conceptualized RS as the disposition to anxiously expect, readily perceive, and overreact to rejection (Downey & Feldman, 1996; Downey, Lebolt, Rincon, & Freitas, 1998; Feldman & Downey, 1994). Because

Geraldine Downey, Antonio L. Freitas, Benjamin Michaelis, and Hala Khouri, Department of Psychology, Columbia University. Antonio L. Freitas is now at the Department of Psychology, Yale University.

This research was supported by grants from the National Institute of Mental Health (R29-MH51113) and the Harry Frank Guggenheim Foundation and by a W. T Grant faculty scholar award.

We thank Rebecca Shansky for her competent help in conducting these studies. We also gratefully acknowledge the contributions of Niatl Bolger, Thomas Bradbury, Carol Dweck, Scott Feldman, Tory Higgins, Sheri Levy, Walter Mischel, and Yuichi Shoda.

Correspondence concerning this article should be addressed to Geral- dine Downey, Department of Psychology, Columbia University, Scher- merhorn Hall, Room 406, Mail Code 5501, 1190 Amsterdam Avenue, New York, New York 10027. Electronic mail may be sent to gdowney @psych.columbia.edu.

545

we view anxious expectations of rejection by significant others as being at the core of RS, we have operationalized RS as anxious expectations of rejection in situations that afford the possibility of rejection by significant others. We refer to people who tend to anxiously expect rejection as high RS (HRS) and those who more calmly expect acceptance as low RS (LRS). Our prior research has documented a link between RS and re- peated experiences of rejection from significant others (Bon- ica & Downey, 1997; Downey, Khouri, & Feldman, 1997; Feld- man & Downey, 1994). Such experiences are thought to lead people to form rejection expectancies that are subsequently activated in situations where rejection is possible.

Once activated, anxious expectations of rejection are thought to prompt a readiness to perceive rejection. Accordingly, in both experimental and field studies, HRS people have been found to perceive rejection in ambiguous cues more readily than LRS people (Downey & Feldman, 1996; Downey et al., 1998). In an experiment, college students who anxiously expected rejection felt more rejected than others when told that a stranger with whom they had just finished a friendly conversation opted not to meet with them a second time (Downey & Feldman, 1996, Study 2). In a prospective field study, students who entered romantic relationships anxiously expecting rejection more readily perceived hurtful intent in their new partners' ambiguous behavior (e.g., being cool and distant; Downey & Feldman, 1996, Study 3 ). When perceived rejection prompts a behavioral overreaction that a significant other finds aversive, it is likely that the significant other will respond in ways that fulfill the HRS person's rejection expectations and that ultimately predict relationship breakup. This article tests this proposition.

Do Self-Fulfi l l ing Prophecies Operate in Close Relat ionships?

Abundant evidence shows that experimentally induced expectations can evoke confirmatory behavior from strangers (for reviews, see Darley & Fazio, 1980; Hilton & Darley, 1991;

546 DOWNEY, FREITAS, MICHAELIS, AND KHOURI

Jussim, 1986, 1991; D. T. Miller & Turnbull, 1986; Snyder, 1992). However, demonstrating expectancy confirmation in ongoing relationships has proven more difficult. Because third variables such as preexisting partner characteristics are potential alternative explanations for apparent expectancy confirmation, the effects of such third variables must be ruled out (Jussim, 199 I). Well-controlled longitudinal investigations of the impact of people' s naturally occurring expectancies on their close relationships are now accumulating. Murray, Holmes, and Griffin (1996a, 1996b) demonstrated that people's idealizations of their romantic partners predicted improvements in their partners' self- images, controlling for the effect of their partners' preexisting self-perceptions. McNulty and Swann(1994) showed that, over time, college students' self-concepts began to conform to their roommates' appraisals of them.

Although Murray et ai. (1996a, 1996b) did not examine this issue, people' s behavior toward their romantic partners probably mediated the relationship between their idealization of their partners and their partners' self-concepts. Similarly, in McNulty and Swann's (1994) study, students' behavior toward their roommates probably helped account for the increasing congruence between the students' self-concepts and their roommates' appraisals of them. Whereas the role of people's behavior in eliciting cognition-congruent responses from significant others has not yet been demonstrated (for a review, see Jussim & Eccles, 1995), researchers have begun to document the impact of people' s relationship cognitions on their behavior toward significant others.

In particular, researchers have focused on establishing the impact of people's relationship cognitions on their behavior during conflicts with romantic partners. This emphasis on conflict is based on findings that conflict behavior has a distinctive utility in predicting important outcomes (e.g., Gottman, 1979, 1993). There is now evidence that relationship cognitions can predict conflict behavior of the type deemed consequential for relationship quality (for a review, see Bradbury & Kamey, 1993). However, links between relationship cognitions and conflict behavior appear to be stronger for women than for men. For example, women's, but not men's, negative attributions for a spouse's behavior have been shown to predict their negative, unconstructive behavior toward the spouse during conflictual discussions (Bradbury, Beach, Fincham, & Nelson, 1996; Brad- bury & Fincham, 1992, 1993; G. E. Miller & Bradbury, 1995). Similarly, doubts about a spouse's psychological availability were stronger predictors of women's than of men's rejecting behavior toward their spouses during problem-solving discussions (Kobak & Hazan, 1991). Thus, conflicts may be more appropriate situations in which to observe the consequences of women's relationship cognitions than those of men.

Does Conflict Lead to Confirmat ion of Reject ion Expectancies?

Of concern to our present study is whether conflict is a suit- able situation in which to examine the hypothesized process through which people's rejection expectancies are fulfilled. Downey (1997) found that HRS people reported feeling more anxious than LRS people in anticipation of conflicts with romantic partners. This conflict-related anxiety did not reflect a gener-

alized anxiety about interactions with romantic partners because no differences in anticipatory anxiety were found for nonconflictual discussions. Thus, by activating HRS people's anxiety, conflicts may potentially trigger the proposed processes leading to rejection.

However, consistent with previously discussed evidence of gender differences in links between relationship cognitions and conflict behaviors, Downey (1997) found that RS was more strongly associated with women's than with men's pessimism about the course and outcome of conflicts with romantic partners. Specifically, RS was a stronger predictor of concern about rejection during conflict, and of feeling lonely and unloved after conflict, in women than in men. Moreover, HRS women were less confident than HRS men that their efforts to resolve conflicts would be successful. These findings imply that HRS women may be more likely to behave in ways that exacerbate conflicts. Consequently, conflicts may be more likely to reveal the hypothesized processes linking rejection expectations with their fulfillment for women than for men.

Current Studies

This research assessed whether people's anxious expectations of rejection prompt them to behave toward romantic partners in ways that elicit rejection and predict breakup. Toward this end, we tested three specific hypotheses in two studies of dating couples.

Study 1 involved a daily-diary study of dating couples and a I-year follow-up. We used the follow-up data to test whether RS predicted breakup (Hypothesis 1 ). This possibility was im- plied by Downey and Feldman's (1996, Study 4) finding that RS predicted partner dissatisfaction, an established predictor of breakup (e.g., Kayser, 1993; Stephen, 1984). Directly showing that RS predicts breakup would underscore the importance of understanding the mechanisms leading to this outcome.

We used daily-diary data from Study 1 to test whether the cycle leading to the fulfillment of rejection expectations is more evident in conflictual than in nonconflictual situations (Hypothe- sis 2). Based on the above review, we expected that conflicts would be more strongly linked with rejection expectancy confirmation in HRS women than in HRS men. In Study 2, we used a behavioral observation paradigm developed by Gottman (1979) to test whether people's conflict behavior mediated the relation between their rejection expectancies and their partners' postconflict rejecting reactions (Hypothesis 3).

Study 1

For 4 weeks, both members of participating couples provided daily records of their relationship-relevant cognitions, affects, behaviors, and conflicts. One-year follow-up data on the status of their relationships were used to test whether RS predicts breakup (Hypothesis 1 ). The daily-diary data were used to test whether conflicts precipitate the hypothesized process where people's rejection expectations elicit confirmatory responses from their partners (Hypothesis 2). Daily-diary designs are well-suited for investigating the effects of naturally occurring conflict on individual couples over time and the differences between couples in response to conflict (Bolger & Schilling,

REJECTION SENSITIVITY AND REJECTION 547

1991). Because in diary designs participants serve as their own

controls the resulting statistical tests are more powerful than those in conventional between-subject designs. Furthermore, the longitudinal nature of diary designs permits more confident in- ferences about causal directionality between variables (e.g., between conflict and relationship satisfaction; Bolger & Zucker- man, 1995).

The daily-diary data allowed us to test whether, on days following conflict, HRS people ' s partners were more likely than LRS people ' s partners to respond in a rejecting manner, indexed by diminished relationship satisfaction and commitment. We used these measures because they are established predictors of breakup (Stephen, 1984) and because we expected that they would elicit rejecting behavior that is evident to the person. As validation, we tested whether these measures predicted breakup and people ' s perceptions of their partners' rejecting behavior, indexed by withdrawal, criticism, and reduced affection. As discussed above, we anticipated that partners' responses to conflict would be more strongly predicted by women ' s rejection expectancies than by those of men.

Relative to LRS people, HRS people were also predicted to view their partners as behaving in a more rejecting way on the day after a conflict. We expected that the relation between peo- p le ' s RS and their perceptions of postconflict partner rejection would be partially mediated by the partners' self-reported relationship satisfaction and commitment.

Method

Sample and Procedure

Dating couples were recruited through announcements posted on the Columbia University campus to participate for pay in a study of romantic relationships. At least one member of each couple was a student at Columbia University. The study was restricted to couples in committed relationships that had been ongoing for at least 6 months and in which both members of the couple lived in New York City. The mean length of relationship was 18.6 months (SD = 14.2). Fifty-eight percent of the men were Caucasian, 20% were Asian American, 7% were African American, 4% were Hispanic, and 11% were from other ethnic backgrounds. The men's mean age was 22 years (SD = 3.7). Fifty-three percent of the women were Caucasian, 34% were Asian American, 5% were Hispanic, 3% were African American, and 5% were from other backgrounds. The women's mean age was 21 years (SD = 2.9). Partici- pants' ethnicities were unrelated to their RS scores.

Couples who completed the study received $50 in compensation. Each member of a couple who expressed interest in the study was mailed a package containing one consent form, five packets, and five return enve- lopes. The first packet was a background questionnaire that included questions about demographic characteristics, RS, dating history, dating patterns, and the current dating relationship. The final four packets each consisted of seven identical, structured questionnaires to be completed at the end of each day for a total of 28 days. Participants were asked to return each week's set of diaries as soon as they were completed. Participants were also asked to complete both the background questionnaire and the daily diaries separately from their partners and to refrain from discussing their responses with their partners.

All couples who completed the study were heterosexual. At least one member of each of 108 couples who had been dating for at least 6 months contacted us to express interest in the study. In 76% of the 108 couples, at least one member completed a background questionnaire (75 couples, 6 additional women, and 1 additional man). In 54% (58/108)

of couples, both members provided background data and at least 2 overlapping weeks of diary data. In fact, all except 4 male and 4 female partners in these 58 couples completed at least 3 weeks of diaries. Time constraint was the primary reason given by couples who did not complete the diary part of the study.

The RS scores of men and women in the 58 couples who provided at least 2 weeks of diary data did not differ from the scores of their counterparts in couples who completed background questionnaires but provided less than 2 weeks of diary data. The diary analyses reported below are based on these 58 couples. Female partners in these 58 couples completed diaries on 94.2% of days and male partners completed diaries on 94.0% of days. Days on which participants did not complete the diary were treated as missing values. Diary completion rate was not associated with RS in either men or women.

About 1 year after the diary study, we attempted to locate all the couples for whom we had background data to establish if they had broken up. Addresses were obtained from several sources including current university directories and records, internet directories, and com- mercial databases. We succeeded in recontacting 53 of the 75 couples. These included 49 of the 58 couples (84%) in which both partners had provided diary data. There were no significant differences in RS between couples who were recontacted and those who were not.

Background Measures

The background questionnaire included the Rejection Sensitivity Questionnaire (RSQ; Downey & Feldman, 1996), global measures of relationship satisfaction and commitment, and demographic questions.

RSQ. The RSQ assesses the anxious-expectations component of RS. A detailed description of the development and validation of the measure is given in Downey and Feldman (1996). t The measure was initially developed from open-ended interviews in which students were asked what they thought would happen and how they would feel in hypothetical situations in which they were requesting something of a significant other, such as a romantic partner, friend, or parent. Answers varied along two dimensions: (a) degree of concern and anxiety about the outcome and (b) expectations of acceptance and rejection. In pilot-testing, responses along the two dimensions did not covary systematically. Consistent with our adoption of an expectancy-value framework (Bandura, 1986), people who both expected rejection and were concerned about this outcome in various interpersonal situations were of theoretical interest.

The final version of the measure consists of 18 hypothetical situations in which rejection by a significant other is possible (e.g., "You ask your friend to do you a big favor" ). For each situation, people are first asked to indicate their degree of concern or anxiety about the outcome of each situation (e.g., "How concerned or anxious would you be over whether or not your friend would want to help you out?") on a 6-point scale ranging from 1 (very unconcerned) to 6 (very concerned). They are then asked to indicate the likelihood that the other person(s) would respond in an accepting fashion (e.g., "I would expect that he/she would willingly agree to help me out" ) on a 6-point scale ranging from 1 (very unlikely) to 6 (very likely). High likelihood of this outcome represents expectations of acceptance, and low likelihood represents expectations of rejection.

Following from our expectancy-value model of anxious expectations of rejection we computed the RSQ scores as follows: A score for each situation was obtained by weighting the expected likelihood of rejection by the degree of anxiety about the outcome of the request. The score for acceptance expectancy was reversed to index rejection expectancy (expectancy of rejection = 7 - expectancy of acceptance ). The reversed score was then multiplied by the score for degree of anxiety or concern.

~The complete measure is available on the World Wide Web at www.columbia.edu / ~ gd20.


A total (cross-situational) RS score for each participant was computed by summing the RS scores for each situation and dividing by the total number of situations.

Downey and Feldman (1996, Study 1 ) showed that the RSQ is a normally distributed measure that taps a relatively enduring and coherent information-processing disposition. The RSQ test-retest reliability was .83 over a 2-3-week period and .78 over a 4-month period. The anxiety and expectations components of the measure were mildly positively related in a large sample ( r = .18, N = 550). Downey and Feldman ( 1996, Study 3 ) provided evidence that RS was not redundant, in terms of its predictive utility, with established personality constructs to which it is conceptually and empirically related. These include measures of introversion, neuroticism, adult attachment style, social anxiety, social avoidance, and self-esteem.

The median and mean RS scores were 8.90 and 8.30 (SD = 2.90, a = .82, range = 2.40-15.40) for women and 7.42 and 8.00 (SD = 3.20, c~ = .88, range = 1.70-16.60) for men. The median and mean RSQ scores for these couples were somewhat lower than the scores reported in Downey and Feldman (1996) for the overall population of college students from which they were recruited (median/mean: men = 9.56/ 9.73; women = 9.44/9.60). However, it is noteworthy that in the Dow- ney and Feldman (1996) sample, people currently in dating relationships were significantly lower in RS than those not currently in a relationship. This finding probably reflects the fact that some HRS people avoid relationships to avoid rejection (Downey, Feldman, & Ayduk, in press).

Although RS is measured continuously, to simplify the analyses we treated it as a dichotomy. People scoring at or above the median (i.e., 8.90 for women and 7.42 for men) were defined as HRS and could be viewed as tending to anxiously expect rejection (scored 1 ). People scoring below the median were defined as LRS and could be viewed as tending to calmly expect acceptance (scored 0). The results reported below were similar whether RS was treated as a dichotomy or as a continuous measure. Partners' RS scores were not significantly correlated (for the continuous measure, r = . 14, p > . 10; for the dichotomous measure, r = - .01, p > .10).

Relationship satisfaction. A scale assessing satisfaction with the relationship was developed by averaging participants' responses to the following three items: (a) " I am satisfied with our relationship," (b) "Our relationship meets my expectations of what a good relationship should be like," and (c) " I could not be happier in our relationship." Participants indicated the extent to which each statement was true of their feelings on an 8-point scale from 0 (not at all true of my feelings) to 7 (completely true of my feelings). The mean for the three-item scale was 5.74 (SD = 1.19, a = .91) for men and 5.47 (SD = 1.65, a = .95) for women. In a pilot study (n = 148 ), scores on this scale correlated .73 (p < .001 ) with relationship satisfaction as assessed by the Dyadic Adjustment Scale (Spanier, 1976).

Relationship commitment. Participants responded to the question "How committed are you to the relationship?" on a 1 (not committed) to 7 (very committed) scale. The mean was 6.51 (SD = 0.83) for men and 6.50 (SD = 0.69) for women.

Diary Measures

The structured daily diary included questions concerning thoughts and feelings about the couples' relationships, partner behavior, and conflict.

Conflict. Each day, participants were asked to indicate whether they had experienced a conflict or disagreement with their romantic partner that day (the occurrence 6f any conflict was coded " 1 " ; the absence of any conflict was coded " 0 " ) . Women reported conflict on 18% of days, and men reported conflict on 16% of days, F(1, 58) = 5.11, p < .05. Members of a couple agreed about whether conflict had occurred on 89% of days. Level of agreement was not affected by people's RS scores. To maximize the independence of data sources, we used women's reports of conflict when examining the effect of women's RS on their

partners, and we used men's reports of conflict when examining the effects of men's RS on their partners. RS was not significantly associated with conflict rates over the diary period (HRS women: M = .21, SD = .41; LRS women: M = .16, SD = .37, F[1, 56] = 1.77, p = .19; HRS men: M = .16, SD = .36; LRS men: M = .16, SD = .37, F[1, 56] = 0.02, p = .89).

Relationship dissatisfaction. Dissatisfaction with the relationship was assessed with the question, "Overall, how would you describe your relationship today?" Participants responded using a 7-point scale ranging from 1 (terrific) to 7 (terrible). The daily mean was 2.32 (SD = 1.15) for men and 2.36 (SD = 1.21) for women, F(1, 58) = 0.01, n s ,

Thoughts of ending the relationship. Participants indicated whether the following statement applied to them that day: " I thought about ending the relationship" (1 = yes, 0 = no). The mean was .04 (SD = .19) for men and .05 (SD = .23) for women, F(1, 58) = 3.42, p < .10. This measure was used to index daily levels of commitment.

Perceptions of partner behavior. Participants checked which one of a set of three accepting and two rejecting behaviors their partners had enacted toward them that day. The accepting behaviors were as follows: "My partner made me feel wanted," "My partner was physically affec- tionate toward me," and "My partner told me s/he loves me." Responses to these items were summed to form an index of accepting behavior (men: M = 2.02, SD = 0.97, a = .57; women: M = 2.17, SD = 0.99, a = .59, F[1, 58] = 0.05, ns). The relatively low reliability of this measure is likely to work against our hypothesis.

One rejecting behavior item indexed passive rejection in the form of withdrawal: "My partner was inattentive and unresponsive toward me" (men: M = .05, SD = .22; women: M = .09, SD = .29, F[1, 58] = 0.40, p < .05). The other item indexed active rejection: "My partner criticized something I said or did" (men: M = .18, SD = .39; women: M = .17, SD = .37, F[1, 58] = 0.18, ns).

One-Year F o l l o w - U p Contac t

The first member of each couple that we contacted was asked whether the couple was still together. Of the 53 couples we contacted, 29% had broken up.

Diary -Da ta Ana ly se s

This study of male-female couples yielded a data set with two levels of analysis. The within-couple level reflects daily variation over time within a couple or within a focal partner in the couple. The between- couple level reflects differences between couples or between focal members of couples. The within-couple level of analysis could be used to estimate each couple's or couple member's reactivity to conflict (e.g., changes in the male partner's satisfaction with or commitment to the relationship following a conflict) as well as the average level of satisfaction or commitment for each couple or member of a couple. The between-couple level of analysis could be used to examine whether couples that included an HRS person differed in these processes from couples that included an LRS person (e.g., whether couples that included an HRS woman differed from couples that included an LRS woman in the impact of conflict on partner satisfaction; whether members of couples that broke up differed from members of couples that stayed together in average daily satisfaction).

The analyses were conducted using a multilevel or hierarchical linear model approach, which permits the simultaneous analysis of within- and between-couple variation (Bolger & Zuckerman, 1995; Bryk & Raudenbush, 1992; Kenny, Kashy, & Bolger, 1998). In contrast, conventional linear models either aggregate across within-couple data, which results in information loss, or conflate within- and between-couple variation, resulting in incorrect tests of significance (see Kenny et al., 1998).


The diary data analyses we conducted addressed two basic types of questions. The main question was whether the relation between variables measured at the daily level (e.g., between conflict and partners' relationship dissatisfaction) differed for HRS and LRS people. More specifically, we were interested in whether the effect of the previous day's independent variable (e.g., conflict) on today's dependent variable (e.g., partner's relationship dissatisfaction) was contingent on the person's RS. Answering this question using a multilevel approach required specifying two equations, including a within- and a between-couple equation. The within-couple equation specifies that values of the dependent variable for a given couple (e.g., partner dissatisfaction) on a given day, St, is predicted by the level of the dependent variable on the previous day, St-i; the level of the independent variable (e.g., conflict) on the previous day, Ct-~ ; and a residual component of the dependent variable, specific to each day, q,. The variable q, is assumed to have a mean of zero and a constant variance across persons and days. The equation is as follows:

St = ao + alSt-t + a2C,_l + qt. (1)

Thus, we assessed the cross-day effect of the independent variable on the dependent variable and included the lagged value of the dependent variable to permit examination of the impact of the independent variable on change in the dependent variable. The purpose of this cross-day approach was to reduce ambiguity about the causal direction of effects. Estimates of a0, al, and a2 were obtained for all couples in the sample.

The between-couple equation specifies that for each couple (i), the effect (a2~) of the independent variable (Ct_l) on the dependent variable (S,) is a function of the RS of the focal person i, RSi, as follows:

a,.i = bo + blRS~ + ei. (2)

Assuming that RS~ is coded 0 for the LRS group and 1 for the HRS group, bo is the mean responsivity to conflict of the LRS group and bl is how many units higher in responsivity to conflict the HRS group is over the LRS group. If we substitute bo + b~RSi + e~ for a2 in Equation 1, then we obtain the following combined equation:

St = ao + alSt-i + boC,_, + bjCt_lRSi + eiCr-i + q,. (3)

Thus, the coefficient b~ can be thought of as a Conflict x RS interaction effect, and we report it as such in the Results section.

The second type of question that we addressed was whether members of couples that subsequently broke up differed in dissatisfaction and commitment on the average day during the diary period from couples that stayed together. As before, a multilevel analysis approach to this question required that we estimate a within- and a between-couple equation. The within-couple equation specifies that a couple-member' s dissatisfaction, for example, on a given day, S,, is a function of his or her mean level across all days, a0, plus a residual component specific to each day, q,:

S, = a0 + q,. (4)

The between-couple equation specifies that members of couples that break up differ in their mean level of dissatisfaction across all days from members of couples that stay together:

aoi = bo + biBreakupi + el. (5)

Assuming that Breakup~ is coded 0 for couples that stayed together and 1 for couples that split up, then bo is the mean dissatisfaction of the nonbreakup group and bt is how many units higher in dissatisfaction the breakup group is over the nonbreakup group.

These analyses were implemented using a modification of the general linear model procedure (PROC GLM) in the SAS statistical package

(SAS Institute, 1989). This procedure can be used to obtain weighted least-squares estimates for multilevel models (see Kenny et al., 1998).

Resul t s

D o e s R S Pred ic t Breakup ?

Because showing that RS predicts breakup would underscore the importance of investigating mediating processes, we began by testing this hypothesis. The prediction was supported. Forty- four percent of couples that included a HRS woman had broken up within a year of the diary study, compared with 15% of couples that included an LRS woman, Xz(1, N = 53) = 5.39, p < .05. Forty-two percent of couples that included an HRS man had broken up in this 1-year period, compared with 15% of couples that included an LRS man, X2(1, N = 53) = 4.59, p < .05. Logistic regression analyses showed that the effect of people ' s RS on breakup remained significant when their partners ' RS, relationship satisfaction, and commitment assessed prior to beginning the diary study were statistically controlled, women: odds ratio = 5.43, /3 = .47, X2(1, N = 53) = 4.06, p < .05; men: odds ratio = 5.38, /3 = .46, XZ(1, N = 53) = 4.20, p < .05.

D o Par tne r Re la t ionsh ip Dissa t i s fac t ion and Thoughts

o f End ing the Re la t ionsh ip Pred ic t Breakup ?

The next analyses were undertaken with daily-diary data to validate that, by assessing the impact of RS on partners ' daily relationship dissatisfaction and thoughts of ending the relationship, we were focusing on outcomes that predicted breakup. Thus, we tested the prediction that the average daily level of relat ionship dissatisfaction and thoughts of ending the relationship, which we viewed as a daily index of relat ionship commitment, in male and female partners predicted breakup over the following year. Multilevel analyses revealed support for the predictions for both men and women.

Men and women in relationships that ended during the 1-year follow-up period were more dissatisfied at the daily level than their counterparts in relationships that remained i n t a c t - - m a l e breakup: M = 2.84, SD = 1.29; male intact: M = 2.12, SD = 1.02, F (1 , 47) = 12.51, p < .001; female breakup: M = 2.73, SD = 1.40; female intact: M = 2.21, SD = 1.07, F (1 , 47) = 5.44, .p < .05. Similarly, the daily probabil i ty of thoughts of ending the relationship was higher in relationships that subsequently ended than in those that continued for both men and w o m e n - - m a l e breakup: M = .08, SD = .27; male intact: M = .02, SD = ,14, F (1 , 47) = 8.33, p < .01; female breakup: M = .13, SD = .34; female intact: M = .03, SD = .17, F (1 , 47) = 23.18, p < .001.

I m p a c t o f Confl ic t on the Par tners o f H R S and L R S

Peop le

Having established that daily relationship satisfaction and commitment dist inguish men and women in relationships that subsequently end from those that remain intact, we then examined processes l inking RS with daily levels of partner satisfaction and commitment. We used the multilevel approach described earlier to examine (a ) whether conflict had a more nega-


tive impact on HRS people's partners than on LRS people's partners and (b) whether HRS and LRS people were aware of their partners' differential responses following conflict. Separate analyses were conducted for men and women.

Does Conflict More Negatively Affect HRS Than LRS People's Partners?

On the day after a conflict, HRS people's partners were expected to experience more relationship dissatisfaction and more thoughts of ending the relationship than LRS people' s partners. Conflict effects were expected to be stronger for HRS women's partners than for HRS men's partners. We examined partners' self-reports of (a) relationship dissatisfaction and (b) thoughts of ending the relationship as a function of their previous day's reports of these variables, previous day's conflict, RS, and the interaction between conflict and RS.

Men. The Conflict X Men's RS interaction was not significant either for partners' relationship dissatisfaction, b = .03, F (1 , 47) = 0.02, p = .86, or for partners' thoughts of ending the relationship, b = .01, F ( 1, 47) = 0.09, p = .76. Thus, HRS and LRS men's partners did not differ significantly in the extent to which conflict affected their relationship satisfaction or commitment.

Women. The hypothesis was supported for women. The re- suits are reported in Table 1 (Models 1 and 2). The partners of HRS and LRS women differed significantly more from each other on days that were not preceded by conflict than on days that were preceded by conflict. This is indicated by significant interactions between conflict and women' s RS both for partners' relationship dissatisfaction, b = .34, F ( 1, 52) = 5.03, p < .05, and for partners' thoughts of ending the relationship, b = .08, F(1 , 52) = 6.22, p < .05. The nonsignificant coefficients for RS in Table 1 (Models 1 and 2) show that partners of HRS and LRS women did not differ significantly from one another in relationship dissatisfaction or in thoughts of ending their relationships on days that were not preceded by conflict.

Figures 1 and 2 give the predicted levels of the current-day's relationship dissatisfaction and thoughts of ending the relationship, respectively, for the partners of HRS and LRS women as a function of whether a conflict had occurred on the previous day. Postconflict differences between HRS and LRS women's partners reflect the following pattern: On days that were preceded by conflict, relative to those that were not, HRS women's partners showed significant increases in relationship dissatisfaction, b = .22, F( 1, 52) = 4.50, p < .05, and in thoughts of ending the relationship, b = .05, F( 1, 52) = 5.90, p < .05. By contrast, LRS women's partners showed nonsignificant declines in dissatisfaction, b = - .12, F( 1, 52) = 0.88, p = .35, and i n thoughts of ending the relationship, b = - .03, F(1 , 52) = 1.05, p = .31.

Are HRS and LRS People Aware of Their Partners' Differential Responses to Conflict?

We expected that, compared with LRS people, HRS people would perceive their partners as being less accepting and more rejecting on days following a conflict. As above, women's RS was expected to have a more pronounced effect than men's RS. Using multilevel analyses, we assessed people's perceptions of their partners' current-day levels of accepting behavior and of rejecting behavior (i.e., withdrawal and criticism) as a function of the previous day' s respective perceived partner behavior, previous day's conflict, RS, and the interaction between the previous day's conflict and RS.

Men. The Conflict x Men's RS interaction was not significant for men 's perceptions of the following behaviors: partners' accepting behavior, b = - .10 , F ( 1, 47) = 0.84, p = .36; withdrawal, b = .05, F ( 1, 47) = 2.50, p = .12; or critical behavior, b = .05, F (1 , 47) = 0 .51,p = .48. Thus, HRS and LRS men's perceptions of their partners' behavior did not differ more on days preceded by a conflict than on days not preceded by a conflict.

Women. The hypothesis was supported for accepting behavior and withdrawal (see Table 1, Models 3 and 4) but not for

Table 1 Impact of Women's Rejection Sensitivity (RS) and Yesterday's Conflict on Their Partner's Feelings About the Relationship and Behavior Today (Based on Multilevel Analysis)

Today's value of dependent variable Intercept Yesterday's value of Yesterday's conflict RS dependent variable (1 = yes, 0 = no) (HRS = 1, LRS = 0)

Yesterday's Conflict x RS

Model 1 Partner's relationship dissatisfaction

(self-report) 1.16 0.48*** -0.12 0.10 0.34* Model 2

Partner's thoughts of ending the relationship (self-report) 0.03 0.22** -0.03 -0.01 0.08*

Model 3 Partner's accepting behavior

(woman's report) 1.27 0.48*** 0.09 -0.23* -0.27* Model 4

Partner's withdrawal (woman's report) 0.06 0.10"* -0.03 0.03 0.09*

Note. HRS = high rejection-sensitive; LRS = low rejection-sensitive. *p < .05. **p < .01. ***p < .001.


2.7

2.6

2.5 =.

2-,

~ ~ 2.3

2.1

2.0

J HRS Women

" " ! LRS Women

t i

No Conflict Conflict

Yesterday Yesterday

Figure 1. Partner's relationship dissatisfaction as a function of conflict and of woman's rejection sensitivity. Predicted values were based on the following equation for a partner at the mean on dissatisfaction yesterday (M = 2.32): Partner Dissatisfaction, = 1.16 + .48 Partner Dissatisfaction,_, - .!2 Conflict,_~ + .10 Rejection Sensitivity + .34 Rejection Sensitivity × Conflict,_,. HRS = high rejection-sensitive; LRS = low rejection-sensitive.

criticism. On days that were preceded by conflict, the differences in HRS and LRS women ' s perceptions of their partners ' behavior were significantly more pronounced than on days that were not preceded by conflict. This is indicated by significant interactions between conflict and women ' s RS for perceptions of partners ' accepting behavior, b = .27, F ( 1, 52) = 5.00, p < .05, and withdrawal, b = .09, F ( 1, 52) = 4.03, p < .05. The interaction was not significant, however, for perceptions of part-

ners ' criticism, b = .08, F ( 1 , 52) = 1.49, p = .23. As the coefficient for RS in Table 1 (Model 4) shows, on days that were not preceded by conflict HRS and LRS women did not differ significantly in their perceptions Of partner withdrawal. However, as the coefficient for RS in Table 1 (Model 3) shows, HRS women perceived their partners as being significantly less accepting than did LRS women even on days that were not preceded by conflict.

0 . 0 9 ' A

_c ~ 0.08,

._c a . • ; 0 . 0 7

. m

o.os

o • 0.04.

• ~ ~ 0.03 " 0 ~ ~, o . ~

o . , 6 0.01

HRS Women

L

~'B LRS Women

No conflict Conflict yesterday yesterday

Figure 2. Partner's likelihood of thinking of ending the relationship as a function of conflict and of woman's rejection sensitivity. Predicted values were based on the following equation for a partner at the mean on thoughts of ending the relationship yesterday (M = .034): Partner Thoughts of Ending Relationshipt = .03 + .22 Partner Thoughts of Ending Relationshipt_~ - .03 Conflict,_m - .01 Rejection Sensitivity + .08 Rejection Sensitivity × Conflict,_1. HRS = high rejection-sensitive; LRS = low rejection-sensitive.

5 5 2 DOWNEY, FREITAS, MICHAELIS, AND KHOURI

W " i . A

f:o' t l l . . Q-O

-=~.

2 . 7 '

2.6"

2.5"

2.4.

2.3

2.2

2.1

2.0

1.9

1.8

LRS Women

HRS Women


Figure 3. Woman's report of partner's accepting behavior as a function of conflict and woman's rejection sensitivity. Predicted values were based on the following equation for a partner at the mean on accepting behavior yesterday (M = 2.24 ): Partner' s Accepting Behavior, = 1.27 + .48 Partner' s Accepting Behavior,_ + .09 Conflictt_t - .23 Rejection Sensitivity - .27 Rejection Sensitivity × Conflict,_]. HRS = high rejection-sensitive; LRS = low rejection-sensitive.

Figures 3 and 4 give the predicted levels of the current-day 's perceived partners ' accepting behavior and withdrawal, respectively, for HRS and LRS women as a function of whether a conflict had occurred on the previous day. Postconflict differences between HRS and LRS women ' s perceptions o f their partners ' behavior reflected the following pattern: On days preceded by conflict, HRS women perceived their partners as being significantly less accepting, b = - . 1 8 , F (1 , 52) = 4.57, p < .05, and more withdrawn, b = .06, F (1 , 52) = 4.02, p < .05,

than on days that were not preceded by conflict. The pattern was reversed for LRS women ' s perceptions of their partners ' behavior but not to a significant degree (accepting behavior: b = .09, F[1, 52] = 0.96, p = .33; withdrawal: b = - . 0 3 , F[1, 52] = 0.72, p = .40).

These findings suggest that the differential self-reported reactions of LRS and HRS women ' s partners to conflict are evident in their next day ' s behavior. We expected this differential effect of conflict on LRS and HRS women ' s perceptions of their part-

0.20

0.18

= A 0.16

e ~ 0.14

"6 ~ o.12

1: ~ 0.10 g~ .~ ~ o.o8 w ~ 0.06

Q. i ~ 0.04

~: 0.02

0.00

HRS Women

LRS Women


Figure 4. Woman's report of partner's withdrawal as a function of conflict and woman's rejection sensitivity. Predicted values were based on the following equation for a partner at the mean on withdrawal yesterday (M = .09): Partner Withdrawal, = .06 + . 10 Partner Withdrawal,_1 - .03 Conflict,_] + .03 Rejection Sensitivity + .09 Rejection Sensitivity × Conflictt_~. HRS = high rejection-sensitive; LRS = low rejection- sensitive.


ners' behavior to be mediated by their partners' dissatisfaction and thoughts of leaving the relationship. To test this hypothesis, we recomputed the multilevel models used to estimate the impact of the prior day's Conflict x Women's RS on women's current-day perception of partners' accepting behavior and withdrawal (see Table 1, Models 3 and 4), controlling in each case for partners' current-day dissatisfaction and thoughts of ending the relationship.

After the mediators were added, the Women's RS × Conflict coefficient reduced to a trend for perceived accepting behavior, b = - .20, F(1, 52) = 3.03, p = .09, and to nonsignificance for perceived withdrawal, b = .05, F(1 , 52) = 1.74, p = .19. For perceived accepting behavior, the b for Women's RS × Conflict declined from - .27 to - .20. Thus, partner dissatisfaction and thoughts of ending the relationship accounted for 26%-- tha t is, [ ( - . 2 7 ) - ( - . 2 0 ) ] / - . 2 7 - - o f the effect of Women's RS × Conflict on women's perception of their partners' accepting behavior. For perceived withdrawal, the b for Women's RS × Conflict declined from .09 to .05. Thus, partner dissatisfaction and thoughts of ending the relationship accounted for 44% of the effect of Women's RS × Conflict on women's perceptions of partner withdrawal.

Additional analyses revealed no significant differences between HRS and LRS women in the association among partners' self-reports of relationship dissatisfaction, thoughts of ending the relationship, and women's perceptions of their partners' rejecting or accepting behavior. This was true when the relevant associations were examined for days that were preceded by conflict as well as for days that were not preceded by conflict.

Discussion

Supporting Hypothesis 1, RS predicted relationship breakup for both men and women, even when controlling for partners' initial level of RS, relationship satisfaction, and commitment. Results from the diary study helped illuminate the processes whereby RS undermines relationships in the case of women but not of men. Specifically, we found that naturally occurring conflicts triggered a process through which women's rejection expectancies led to their partners' rejecting responses, operationalized as partner-reported relationship dissatisfaction and thoughts of ending the relationship. Both of these indexes of rejection predicted breakup for men and for women.

On days preceded by conflict, HRS women's partners were more likely than LRS women's partners to experience relationship dissatisfaction and to think of ending the relationship. Moreover, HRS women's partners felt significantly more negative about their relationships on days preceded by conflict than on days that were not. The pattern was reversed, but to a nonsignificant degree, for the partners of LRS women. These findings are not attributable to the stable effects of partner background characteristics, because these were held constant in within-couple analyses. Nor can they be accounted for by the contaminat- ing effect of prior day's dissatisfaction and thoughts of ending the relationship, which were also held constant in the analyses.

The differential impact of conflict on the partners of HRS and LRS women was evident to the women. On days preceded by conflict, partners were perceived to be less accepting and

more withdrawn by HRS than by LRS women. This link was only partially mediated by partner satisfaction and commitment.

Conflicts did not precipitate changes in relationship satisfaction or commitment for HRS and LRS men's partners. Thus, these findings add to accumulating evidence that typical conflicts may be more appropriate situations for examining the impact of women's rather than of men's relationship cognitions.

Overall, Study 1 findings implicate conflicts as critical situations in which to examine the processes leading to the fulfillment of HRS women's rejection expectations. However, there are a number of limitations to the diary approach that qualify the confidence with which such conclusions can be drawn. First, partners' relationship perceptions were not assessed immedi- ately before and after the conflict. Thus, rather than resulting from the conflict per se, the more rejecting responses of HRS women's partners on the day after a conflict may have been caused by the negative interactional residue of a poorly resolved conflict. Second, the daily-diary approach relies on people com- pleting self-reports each day. Poor compliance may have dis- torted the study results.

Study 2

The first goal of Study 2 was to address the limitations of the diary approach by investigating under more controlled con- ditions whether conflict had a differential impact on HRS and LRS women's partners. The second goal was to determine whether the negative behavior of HRS women during a conflict would account for their partners' postconflict rejecting responses (Hypothesis 3). In this study, partners' postconflict anger about their relationships was used as an index of rejection because anger is a common reaction to negative interactions and because it has been shown to predict rejecting behavior (e.g., Juvonen, 1991).

To replicate the Study 1 finding that conflict would precipitate rejection by HRS women's partners, we first examined whether HRS women's partners would feel more angry about their relationships following a conflict discussion than would LRS women's partners. Prior evidence that laboratory discussions of conflictual issues improved romantic partners' moods (Bradbury, 1994; Bradbury & Davila, 1997; Veroff, Hatchett, & Douvan, 1992) led us to expect an average pre- to postconflict decline in partner anger. This decline was expected to be significantly more pronounced in LRS women's partners than in HRS women's partners.

Next, we tested whether, during the conflict discussion, HRS women would behave in a more negative way toward their partners than would LRS women. Finally, we tested whether differences in HRS and LRS women' s negative conflict behavior would help account for postconflict differences in their partners' anger levels. On the basis of our findings in Study 1, these predictions were not expected to be supported for men.

Following Jussim ( 1991 ), we sought to rule out third-variable explanations for the hypothesized impact of the person's anxious expectations of rejection on the partner's postconflict anger and on the person's own conflict behavior. Potential third-variable explanations included the person's preconflict mood and the partner's own level of RS, current relationship satisfaction and commitment, preconflict anger, and conflict behavior. Thus, we


tes ted whether contro l l ing for these variables al tered suppor t for the behav io r media t ional model .

Method

Sample

Participants were 39 college-age couples in the early stages of dating relationships. The couples had been dating for an average of 3.3 months (SD = 1.4), with a minimum of 2 months and a maximum of 6 months. The mean age of both men and women was 20 years. At least 1 member of each couple was a university student. The racial composition of the sample was as follows: 67.4% of the men were Caucasian, 2.2% were African American, 8.7% were Hispanic, 15.2% were Asian American, and 6.5% were from other backgrounds; 57.8% of the women were Caucasian, 2.2% were African American, 2.2% were Hispanic, 22.2% were Asian American, and 15.6% were from other backgrounds.

The study included only couples in which both members reported that the relationship was exclusive. This criterion was used to preclude the possibility that people might feel rejected by and negative towards their partners because of their partners' other romantic involvements.

Procedure

Flyers were posted around the Columbia University campus and in students' mailboxes inviting couples to participate for pay in a study of dating relationships. All participants signed consent forms that included assurances of confidentiality and of the participants' freedom to drop out of the study at any point. Each member of the couple separately completed a background questionnaire that included measures of RS, relationship satisfaction and commitment, exclusivity of the relationship, and demographic information. Forty-eight couples completed the background questionnaires.

One to 2 weeks later, couples came into the laboratory to be videotaped discussing an unresolved relationship issue. On arrival, both members separately completed a measure of their current mood when thinking of their partner and of the relationship.

Participants were then given a list of 19 topics that college dating couples report arguing about (see below). They were asked (a) to choose up to 5 topics of ongoing conflict to discuss during the video session and (b) to indicate the most salient issue. This approach to selecting discussion topics was based on Gottman's (1979) finding that couples become more involved when discussing topics of personal concern than when discussing standard experimenter-assigned topics. The list had been generated from the most common answers to the open- ended question, "What are some of the things you and your dating partner argue about?", provided by 100 participants in a pilot study of college students. These items included "spending time together, . . . . other friendships," "commitment," and "sex."

The researcher then picked a topic that both partners had selected to be the focus of their discussion. In the four couples whose members selected nonoverlapping topics, the researcher randomly selected a topic that one member of the couple had picked as especially salient. HRS and LRS people did not differ in their selection of topics.

Couples were filmed for 20 min. The video camera was set up behind a one-way mirror to reduce distraction. Participants were aware that they were being recorded on video but they were assured that nobody would be watching them during the interaction period. After 20 rain, the camera was turned off and the interaction ended. To assess postdiscussion mood, each partner again completed the mood questionnaire. When finished, participants were asked for a second time for consent to use the video for research purposes and were given the opportunity to question the experimenter about the study. Each participant received $14 in compensation for his or her time.

The laboratory interaction component was completed successfully by 39 couples. Of the remaining 9 couples who had completed the background questionnaire, 3 couples had broken up before being videotaped, 3 couples could not be scheduled for videotaping before leaving campus for the summer, and 2 couples declined to participate in the videotaped discussion. One couple completed the conflict discussion, but their data could not be used because of technical difficulties. The 39 couples on whom interaction data were obtained did not differ significantly from the original 48 couples on RS or on relationship satisfaction or commitment. An additional 23 couples expressed an interest in participating in the study but did not complete any measures.

The sample included two lesbian couples. These couples were ex- cluded from the analyses examining the impact of men's RS. For the analyses of the impact of women's RS, one member of each lesbian couple was randomly assigned to be the focal woman. Dropping these two couples from the analyses did not alter the findings reported below.

Measures

RSQ. The RSQ was described in Study 1. The mean for women was 8.1 (SD = 3.6), and the median was 8.3. The mean for men was 8.8 (SD = 3.2), and the median was 8.0. As in Study 1, the mean and median RSQ scores for these couples were somewhat lower than the scores reported in Downey and Feldman (1996) for the overall population of college students from which they were recruited. As in Study 1, we treated RS as a dichotomy. People scoring at or above the median were defined as HRS (scored 1), whereas those scoring below the median were defined as LRS (scored 0). Similar results emerged whether RS was treated dichotomously or continuously. Partners' RS scores were not significantly correlated (for the continuous measure: r = - .11 , p > .10; for the dichotomous measure, r = - .15, p > .10).

Relationship satisfaction. This measure was described in Study 1. The mean on a 0 - 7 scale was 5.31 (SD = 1.29, ct = .79) for men and 5.45 (SD = 1.30, ct = .91) for women.

Relationship commitment. As in Study 1, commitment was indexed by responses on a 1 (not committed) to 7 (very committed) scale to the question, "How committed are you to the relationship?" The mean was 6.34 (SD = 0.84) for men and 6.03 (SD = 0.96) for women.

Mood questionnaire. This questionnaire was used to assess anger about the relationship, which we viewed as an indicator of the partners' rejecting response following conflict. The questionnaire also assessed depression and anxiety about the relationship, the preconflict measures of which were included as covariates in some of the analyses reported below. Participants were asked to rate, on a 0 - 3 scale, how they felt "right now" when they thought of their partner and their relationship. The mood items were drawn from the Affects Balance Scale (Derogatis, 1975).

The Anger scale consists of feeling resentful, irritated, frustrated, enraged, wary, threatened, and angry at the other person. The Depression scale consists of feeling guilty, hopeless, sad, worthless, depressed, blue, ashamed, and unhappy. The Anxiety scale consists of feeling afraid, nervous, agitated, tense, on edge, and anxious. Alpha reliabilities for Anger, Depression, and Anxiety were respectively .78, .82, and .77 for men and .78, .80, and .65 for women.

Interaction data. Videotaped interactions were coded using the Mar- ital Interaction Coding Sys tem-- IV (MICS-IV; Weiss & Summers, 1983). To ensure that participants had become comfortable in the laboratory setting, only the final 10 min of the taped session were coded. Coding was done by experienced coders at the University of Oregon Marital Studies Program (for a discussion of the MICS-IV coding scheme, see Weiss & Summers, 1983). This microanalytic coding scheme involves coding each discrete segment of classifiable action that participants engage in during the taped session. The MICS- IV codes encompass both verbal content and nonverbal behavior. HRS and LRS people engaged in the same overall amounts of behavior.


We combined a set of behavior codes to form a negative-behavior composite and calculated the proportion of total behavior that was negative. The codes that made up the negative-behavior composite and their definitions given in the MICS-IV manual were as follows: mindread negative (a statement of fact that assumes a negative mindset or motivation of the partner), voice tone (indicates a hostile or negative voice tone), deny responsibility (a statement that conveys lack of responsibility for a problem), put-down (a verbal statement or nonverbal behavior that demeans and mocks the partner), turn-off (nonverbal gestures that communicate disgust, displeasure, disapproval or disagreement), and dysphoric affect (an affect that communicates depression or sadness; any self-complaint or whiny voice). No participants engaged in two additional types of negative behaviors coded by the MICS: criticize and threat. The negative-behavior composite accounted for 2% of the total number of behaviors coded for women and 1% for men. The relatively low rates of negative behaviors were to be expected given that the sample was nondistressed and given that a high proportion of all the coded behaviors were statements relating to the problem being discussed (35 % ) and inaudible talk (11%) (R. L. Weiss, personal communication). Cod- ers were unaware of the study hypotheses or of participants' RS scores. Two people coded 20% of the tapes. The interrater agreement between the two coders was 75%.

Results

Because the behavior data were proportional, analyses were performed on behavior data that had been transformed using the arcsin square root transformation (Myers, 1966); means and standard deviations are reported on the basis of untransformed behavior data.

Men

The behavioral mediation hypothesis was not supported for men. The results of an analysis of covariance (ANCOVA) showed that partners of HRS men were not significantly more angry about the relationship following the conflict discussion than partners of LRS men, adjusting for preconflict partner anger (HRS: adjusted M = .25; LRS: adjusted M = .23, t[34] = 0.17, p = .87). Men's RS also was not significantly associated with partners' postconflict depression and anxiety. HRS and LRS men did not differ significantly in the negativity of their behavior during the conflict discussion (HRS: M = .02, SD = .02; LRS: M = .01, SD = .02, t[35] = 0.92, p = .36).

Women

The behavioral mediation hypothesis was supported for women. The results of an analysis of variance (ANOVA) showed that partners of HRS women were significantly more angry about the relationship following the conflict discussion than partners of LRS women (HRS: M = .34, SD = .43; LRS: M = .11, SD = .20, t[37] = 2.15, p < .05). HRS and LRS women' s partners did not differ significantly in preconflict anger (HRS: M = .30, SD = .40; LRS: M = .20, SD = .39, t[37] = 0.83, p = .41 ). The postconflict difference remained significant when partners' preconflict anger was included as a covariate in an ANCOVA (HRS: adjusted M = .30; LRS: adjusted M = .14, t[36] = 2.14, p < .05). Adjusting for preconflict anger, the postconflict difference between HRS and LRS women in their partners' anger reflected both a significant pre- to postconflict

discussion decline in anger in partners of LRS women, M = - .11 , t (36) = 2.10, p < .05, and a nonsignificant increase in anger in partners of HRS women, M = .05, t (36) = 0.96, p = .35. For the sample as a whole, there was a nonsignificant decline in anger from pre- to postconflict, M = - .03 , SD = .29, t (37) = 0.70, p = .49. The effect of women's RS on their partners' postconflict mood was specific to anger. Women's RS was not a significant predictor of their partners' postconflict depression or anxiety.

An ANOVA also revealed that HRS women showed propor- tionately more negative behavior than LRS women during the discussion (HRS: M = .04, SD = .04; LRS: M = .01, SD = .02; t[37] = 2.97, p < .001 ). This difference remained significant when partners' preconflict anger was entered as a covariate, t (36) = 2.81, p < .01.

To assess how much of the association between women's RS and their partners' postconflict anger was explained by the women's negative behavior during the conflict, we conducted a path analysis. First, we regressed the partners' postconflict anger on their preconflict anger and on the women's RS. The b for RS was .24, t (37) = 2.14, p < .05. When women's negativity during the interaction was controlled, the b for the women's RS declined from .24 to .11. This latter coefficient is the direct effect of the women's RS on their partners' change in anger. The indirect effect of women's RS on their partners' change in anger is .24 - . 11 = . 13. Thus, the women' s negative behaviors accounted for 54% ( .13/ .24) of the effect of women's RS on their partners' change in anger. These results are presented in Figure 5.

We recomputed the various analyses with a number of control variables to rule out potential third-variable explanations for the impact of women's RS on their behavior and on their partners' postconflict anger. Controlling for partners' RS, global relationship satisfaction or commitment, or their negative behavior during their interactions did not alter the results. Similarly, the results were not altered by controlling for the women's relationship satisfaction or commitment, or preinteraction negative mood. Finally, the effect of negative behavior on partner postinteraction anger was similar for LRS and HRS women (RS × Women's Negative Behavior: b = .31, t[34] = 1.17, p = .25).

Discussion

These results support the behavioral mediation hypothesis (Hypothesis 3) for women but not for men. First, although the partners of HRS and LRS women did not differ significantly in preconflict anger, the partners of HRS women were significantly more angry about their relationships after conf ic t than were the partners of LRS women. These postconflict differences reflected a nonsignificant increase in the anger of HRS women's partners and a significant decline in the anger of LRS women's partners, which is consistent with prior research (Bradbury, 1994; Brad- bury & Davila, 1997; Veroff, Hatchett, & Douvan, 1992).

Second, HRS women behaved more negatively than LRS women during the conflict discussion. Third, HRS women's greater negativity during their discussions helped account for why their partners were more angry than LRS women's partners after their discussions. These results held irrespective of the women's preinteraction negative mood, relationship satisfaction


Woman's Rejection Sensitivity

Woman's Negative Behavior During Interaction

.11 d

.48 * ~

Partner's Post-interaction Anger

Figure 5. Does women's negative behavior mediate the impact of their rejection sensitivity on their partner's postinteraction anger? Analyses controlled for partner's preconflict anger. Total effect of women's rejection sensitivity on partner's postinteraction anger was .24*. *p < .05; * *p < .01.

and commitment, their partners' relationship satisfaction and commitment, RS, preinteraction negative mood, and negative behavior during the interaction. Consistent with Study l ' s findings, men' s RS predicted neither their conflict behavior nor their partners' postconflict anger.

General Discuss ion

Our two studies join other recent studies (McNulty &Swann, 1994; Murray et al., 1996a, 1996b) in answering Jussim's call for evidence that people's expectations influence, rather than merely reflect, the reality of their ongoing relationships (Jussim, 1991; Jussim & Eccles, 1995). Moreover, our research moves beyond documenting expectancy confirmation and into delineating the intermediary behavioral processes. Specifically, our re- suits provide evidence for Sroufe's (1990) proposal, derived from attachment theory, that rejection expectations can lead people to behave in ways that elicit rejection from others. Using research designs that permitted us to rule out plausible alternative hypotheses for the relation between rejection expectancies and their confirmation, we showed that women who anxiously expected rejection behaved in ways during conflict that elicited a rejecting response from their romantic partners.

Contributions of the Research

Below, we outline the contribution of our findings to the literatures on relationship beliefs in close relationships and on self-fulfilling prophecies in ongoing relationships.

Role of Relationship Beliefs in Close Relationships

Our studies elucidate a process whereby one person's relationship beliefs, presumably formed on the basis of that person's prior relational experience, can lead a couple to become locked into destructive interactional patterns. Our results also point to a process whereby a person's maladaptive relationship beliefs can be maintained, whatever their initial origin. Moreover, our findings suggest how one person's relationship history could help shape the quality of the partner's experiences in subsequent relationships. Specifically, the destructive or healthy interactional processes set in motion by one person's relationship beliefs may alter the relationship beliefs of that person's partner in ways that influence the partner's subsequent relationships.

These possibilities need to be tested directly in subsequent research.

The particular relationship beliefs that we focused on, beliefs about whether significant others will be supportive or rejecting in times of need, are widely acknowledged to shape and be shaped by relationship experiences (e.g., Bowlby, 1969, 1973, 1980; Hazan & Shaver, 1994). An objective of our work on rejection sensitivity was to provide a precise and, consequently, testable account of these beliefs and of their role in people's relationships. Guided by social-cognitive theory, we conceptualized the legacy of rejection in terms of the moment-to-moment thoughts, feelings, and action plans that are the immediate ante- cedents of social interaction (e.g., Andersen & Glassman, 1996; Baldwin, 1992; Dweck & Leggett, 1988; Heider, 1958; Hig- gins & Bargh, 1987; Kelley, 1979; Mischel, 1973; Mischel & Shoda, 1995). Specifically, we conceptualized rejection sensitivity as the disposition to anxiously expect, readily perceive, and overreact to rejection from significant others. Our findings underscore the value of merging our social-cognitive perspective on what individuals bring to and take from relationships with the interpersonal perspective on relationships (Christensen & Heavey, 1990; Coyne, 1976; Gottman, 1979), which emphasizes the social interaction occurring in relationships. Whereas influ- ential theories of social behavior often assume an integrated perspective, it has not been typical to investigate the dynamic interplay between each person's relationship-relevant thoughts, feelings, and goals and the interaction between members of the couple. Yet, failure to do so potentially limits our understanding of romantic relationships.

Self-Fulfilling Prophecy in Naturally Occurring Relationships

A review of existing studies of self-fulfilling prophecy in ongoing relationships led Jussim (1991) to conclude that the evidence is weaker and less consistent than commonly assumed (see also Jussim & Eccles, 1995). Results from our diary study support an explanation for the equivocal nature of this evidence that is consistent with Mischel and Shoda's (1995) explanation for why personality dispositions are generally equivocal predictors of behavior. Mischel and Shoda (1995) have proposed that whether an individual's personality disposition (e.g., RS) becomes activ.ated in ways that influence behavior depends reli-


ably on the psychological meaning of the situations that the individual encounters. Thus, for example, situations where rejection is perceived to be a possibility (e.g., conflict) should activate anxious expectations of rejection in HRS women to a greater extent than the average situation. Consequently, Mischel and Shoda's (1995) dynamic account of personality predicts a pattern of weak and inconsistent results from studies of expectancy confirmation that aggregate across expectancy-relevant and expectancy-irrelevant situations.

The diary methodology allowed us to capitalize on naturally occurring, within-person variation in situations deemed likely to activate rejection expectancies (i.e., conflicts) and to compare expectancy realization in these situations with expectancy realization in situations unselected for their expectancy-activating potential. There was no evidence of expectancy realization in nonconflict situations, which we view as proxying for the aggre- gation approach in that no distinction is made between expectancy-relevant and expectancy-irrelevant nonconflict situations. However, expectancy realization occurred in the specific context of conflict, which was known to activate rejection expectancies in HRS women. These findings highlight the importance of expectancy-relevant situations in revealing expectancy realization. By implication, studies of self-fulfilling prophecy hypotheses in ongoing relationships can yield evidence of expectancy confirmation only to the extent that participants are exposed to situations that activate their pertinent expectations. Research on self-fulfilling prophecy in ongoing relationships generally has not attempted to isolate key situations in which to examine expectancy confirmation (for exception, see Madon, Jussim, & Eccles, 1997). However, research from both an attribution and an attachment perspective provides several excellent illustrations of the value of carefully selecting relevant contexts in which to investigate the implications of people's relationship cognitions for their interpersonal behavior (e.g., Bradbury & Fincham, 1992; Kobak & Hazan, 1991; Simpson, Rholes, & Nelligan, 1992). Particularly pertinent is a recent study by Simpson, Rholes, and Phillips (1996) in which dating couples were observed when discussing either a major or minor relationship problem. Differences in the negativity of conflict behavior between women who were high and low in ambivalent attachment were evident only during the discussion of a major relationship problem.

Issues f o r Future Research

Although answering the questions of whether and why rejection expectancies are realized, our findings also raised important questions.

What explains the gender differences in our findings? Whereas our results reveal the potential of interpersonal conflict as a critical situation for observing the realization of rejection expectancies, they also delineate its boundaries. Conflicts with romantic partners were appropriate situations in which to examine the confirmation of rejection expectations in women but not in men. Perhaps gender differences in the meaning of conflicts and disagreement with intimate partners account for this finding. Consistent with this possibility, the prospect of disagreement and conflict has been found to elicit stronger rejection concerns in HRS women than in HRS men (Downey, 1997). Similarly,

adult attachment research has shown links between relationship arguments and abandonment anxiety for women but not for men (Collins & Read, 1990).

Why might intimate conflicts have different meanings for men and women? Cross and Madson (1997) recently proposed that maintaining harmonious intimate relationships is a more integral part of the self-concept in women than in men. If that is true, then one implication is that events that threaten closeness and connection with significant others, including private conflicts and disagreement, should activate rejection concerns in HRS women to a greater extent than in HRS men. What type of events may signify rejection to HRS men to a greater extent than to HRS women? Baumeister and Sommer (1997) have argued that whereas connection with significant others may be relatively more integral to the self-concepts of women than of men, the status and esteem in which men are held by the larger social group may be relatively more integral to men's self- concepts. Thus, HRS men may be particularly likely to perceive rejection in events that threaten the loss of societal respect or in events that challenge their confidence that others' respect them (e.g., if their partners make them look foolish in front of their friends or undermine their belief that others respect them). This speculation is consistent with Downey and Feldman's (1996) finding that HRS men, but not HRS women, were char- acterized by sexual jealousy and by efforts to monitor and control their partners' social contacts. These behaviors can readily be interpreted as reflecting attempts to control a partner's behavior to maintain the respect of one's social group. Not surpris- ingly, such behavior contributed to partner dissatisfaction.

These observations suggest that the relation between women's RS and their rejection-eliciting hostility should be particularly evident in situations that activate concerns about loss of a close, intimate relationship. In contrast, the relation between men's RS and their rejection-eliciting jealous and controlling behavior should be evident in situations that activate concerns about loss of social status. Research that tests these explanations could help illuminate mechanisms underlying our finding that men's rejection sensitivity predicts relationship breakup.

Why do HRS women behave in ways that elicit rejection? By focusing in Study 2 on a situation selected to activate rejection expectancies (i.e., conflict), it was possible to identify behaviors that led to the confirmation of women's rejection expectancies. The question of why HRS women engaged in rejection-eliciting behavior was not addressed, however. In fact, it appears counterintuitive that women who anxiously expect rejection would behave in ways that are likely to elicit rejection. A possible explanation suggested by self-verification theory (Swann, 1983) is that HRS women, like everyone else, are motivated to maintain predictability and thus a sense of control over their lives. Thus, the goal of maintaining predictability might prompt women who anxiously expect rejection to act in ways that help ensure partner rejection. One way of achieving this goal is to select partners who are likely to act rejectingly, such as someone who is unlikely to show sustained relationship satisfaction or commitment. Yet, our findings cannot be completely explained by HRS women selecting partners who are prone to behave rejectingly. The longitudinal nature of the diary and observational studies made it possible to statistically adjust


for pertinent preexisting partner characteristics, such as satisfaction and commitment.

Although partner selection cannot completely explain why women with anxious expectations of rejection tend to get rejected, the desire to maintain predictability may still motivate women who anxiously expect rejection to behave in rejection- eliciting ways during conflict. However, Ayduk, Downey, Testa, Yen, and Shoda (in press) have found evidence that is more consistent with viewing the hostile behavior of women who anxiously expect rejection as an overreaction to perceived rejection (see also, Collins, 1996). In an experiment using a priming paradigm (Bargh, Raymond, Pryor, & Strack, 1995), thoughts of rejection automatically facilitated thoughts of hostility in HRS women to a greater extent than in LRS women. By contrast, hostile thoughts did not facilitate thoughts of rejection, as should have occurred if HRS women's hostile behavior was intended to elicit rejection. Ayduk et al. (in press) also reported evidence from an experiment and from the diary study reported in Study 1 that HRS women react in a more hostile way than LRS women to similar levels of actual and perceived rejection. Dispositional differences in hostility cannot account for these findings because HRS and LRS women are similar in hostility when not primed by rejecting cues. These findings are consistent with evidence that women's hostility toward significant others often follows perceived threats to their relationships (Harris, 1993). The results also suggest that, rather than having an ex- plicit instrumental goal, HRS women's aggression is probably a behavioral expression of feelings of hurt, anger, despair, and hopelessness (Ben-David, 1993; Eskin & Kravitz, 1980).

Are there alternative responses to rejection cues? Above, we suggested that HRS women tend to interpret rejection cues as evidence that rejection has occurred. Thus, feeling that they have nothing left to lose, their hostility may be an expression of revenge or frustration. An alternative response of HRS women in situations where they fear rejection may be to interpret rejection cues as triggers to engage in behaviors such as self-silencing (Jack, 1991, Jack & Dill, 1992)'or acting ingratiatingly (Dow- ney, Bonica, & Rincon, in press), which are intended to prevent rejection. Both of these forms of behavior may help them prevent overt rejection in the short run. But, in the long run such behavior in HRS women may lead to both depression and diminished relationship satisfaction for both partners as problems in the relationship never get constructively resolved.

What do partners' relationship schema contribute to the rejection-expectancy confirmation process? The process that we have outlined presumes a romantic partner who responds in a typical rejecting fashion to interpersonal negativity during conflict. However, evidence from the broader literature on expectancy confirmation indicates that naturally occurring or experimentally induced partner cognitions and affects may moderate the likelihood that the partner will confirm a person's expectancies (for a review, see Snyder, 1992). Thus, an important goal of future research is to delineate partner qualities that moderate the rejection-sensitivity cycle. What relationship goals and cognitions would motivate and enable a partner to make extra efforts to accommodate the HRS person (see Reis & Patrick, 1996)? What goals and cognitions would accelerate conflict-rejection spirals?

What are the consequences of violated expectations? Our

research focused on demonstrating that rejection expectancies are fulfilled by prompting behavior that elicits confirmatory social feedback. By implication, altering people's rejection expectations should alter their behavior and consequently alter the quality of feedback they receive. In support of this inference, Rabiner and Coie (1989) found that when peer-rejected children were led to expect acceptance from peers they were about to meet, these children were subsequently better accepted than peer-rejected children not led to expect acceptance, and that the girls (but not boys) induced to expect acceptance were observed to behave more positively toward their peers. These experimental results convey the importance of delineating the naturally occurring circumstances that alter people's rejection expectations.

Yet, the malleability of people's expectations about relationships undoubtedly depends on how ingrained and generalized their expectations are and on the goals served by maintaining consistency (Swann, 1997). Generalized expectations of rejection that are based on prolonged childhood rejection by parents and peers are probably less responsive to contradictory feedback than specific expectations about romantic relationships formed on the basis of rejection in a recent relationship (Bowlby, 1969).

Conclusions and Caveats

Several caveats should be noted when interpreting the results of Studies 1 and 2. First, we examined partners' rejecting thoughts and feelings as distinct from observable rejecting behavior. Findings that both relationship breakup and people's perception of their partners' rejecting behavior were predicted by partner satisfaction and commitment validated using these measures to index partner rejection. Nonetheless, there is a need to test explicitly whether RS leads people to behave in ways that elicit observable rejecting behavior from their partners. Second, both studies were restricted to college dating couples in committed and satisfying relationships. Indeed, very little negativity was expressed during the Study 2 conflict discussion. We examined only committed relationships to help ensure that RS was an antecedent rather than a consequence of relationship difficulties. Future research will need to establish whether our findings generalize to distressed relationships. The applicability of our findings to marital relationships also needs to be examined.

Third, the findings suggest that HRS women's anxious expectations of rejection lead to relationship breakup in part because HRS women behave in ways that erode their partners' relationship satisfaction and commitment. Alternatively, as perceived rejections accumulate, HRS women may become increasingly dissatisfied with the relationship, prompting them to end it. Or, having become convinced that their partner will inevitably leave them, HRS women may act preemptively to end the anxiety- laden waiting period and to avoid overt rejection. Our studies did not examine the precise causes of the breakups or who initiated them. A more complete understanding of the relation between RS and relationship longevity requires following people throughout a relationship and recording the breakup process.

These caveats notwithstanding, our results confirm that women's expectancies help create their own reality in romantic relationships. During conflicts, women's expectations of rejection


led them to behave in ways that elicited confirmatory reactions from their romantic partners. Moreover, even when controlling for a partner's relationship satisfaction and commitment, rejection sensitivity proved a potent predictor of relationship breakup. Eventually, then, as conflicts accumulate, the realities of HRS women 's relationships may merge more closely with their expectations.

Re fe rences

Andersen, S., & Glassman, N. (1996). Responding to significant others when they are not there: Effects on interpersonal inference, motivation, and affect. In R. Sorrentino & E.T. Higgins (Eds.), Handbook of motivation and cognition: Vol. 3. The interpersonal context (pp. 262- 321 ). New York: Guilford Press.

Ayduk, O., Downey, G., Testa, A., Yen, Y., & Shoda, Y. (in press). Does rejection elicit hostility in rejection-sensitive women? Social Cognition.

Baldwin, M. W. (1992). Relational schemas and the processing of social information. Psychological Bulletin, 112, 461-484.

Bandura, A. (1986). Social foundations of thought and action. Engle- wood Cliffs, NJ: Prentice Hall.

Bargh, J. A:, Raymond, P., Pryor, J. B., & Strack, F. (1995). Attrac- tiveness of the underling: An automatic power ~ sex association and its consequences for sexual harassment and aggression. Journal of Personality and Social Psychology, 68, 768-781.

Baumeister, R., & Sommer, K. L. (1997). What do men want? Gender differences and two spheres of belongingness: Comment on Cross and Madson (1997). Psychological Bulletin, 122, 38-44.

Ben-David, S. (1993). The two facets of female violence: The public and the domestic domains. Journal of Family Violence, 8, 345-359.

Bolger, N., & Schilling, E.A. (1991). Personality and problems of everyday life: The role of neuroticism in exposure and reactivity to daily stressors. Journal of Personality, 59, 355-386.

Bolger, N., & Zuckerman, A. (1995). A framework for studying personality in the stress process. Journal of Personality and Social Psychol- ogy, 69, 890-902.

Bonica, C., & Downey, G. (1997). Peer rejection sensitizes children to expect rejection. Unpublished manuscript. Columbia University, New York.

Bowlby, J. ~ 1969). Attachment and loss: Vol. 1. Attachment. New York: Basic Books.

Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation. New York: Basic Books.

Bowlby, J. (1980). Attachment and loss: Vol. 3. Loss, sadness, and depression. New York: Basic Books.

Bradbury, T N. (1994). Unintended effects of marital research on marital relationships. Journal of Family Psychology, 8, 187-201.

Bradbury, T. N., Beach, S. R., Fincham, E D., & Nelson, G. M. (1996). Attributions and behavior in functional and dysfunctional marriages. Journal of Consulting and Clinical Psychology, 64, 569-576.

Bradbury, T. N., & Davila, J. (1997). Husband and wife affect before and after conflict discussions. Unpublished data. University of Califor- nia, Los Angeles.

Bradbury, T.N., & Fincham, E D. (1990). Attributions in marriage: Review and critique. Psychological Bulletin, 107, 3-33.

Bradbury, T N., & Fincham, E D. (1992). Attributions and behavior in marital interactions. Journal of Personality and Social Psychology, 63, 613-628.

Bradbury, T. N., & Fincham, E D. ( 1993 ). Assessing dysfunctional cognitions in marriage: A reconsideration of the Relationship Belief In- ventory. Psychological Assessment, 5, 92-101.

Bradbury, T N., & Karney, B. R. (1993). Longitudinal study of marital

interaction and dysfunction: Review and analysis. Clinical Psychology Review, 13, 15-27.

Bryk, A., & Raudenbush, S.W. (1992). Hierarchical linear models. Newbury Park, CA: Sage.

Christensen, A., & Heavey, C. L. (1990). Gender and social structure in the demand/withdraw pattern of marital interaction. Journal of Personality and Social Psychology, 59, 73-81.

Collins, N. (1996). Working models of attachment: Implications for explanations, emotion, and behavior. Journal of Personality and So- cial Psychology, 71, 810-832.

Collins, N., & Read, S. (1990). Adult attachment, working models, and relationship quality in dating couples. Journal of Personality and Social Psychology, 58, 644-663.

Coyne, J. C. (1976). Depression and the response of others. Journal of Abnormal Psychology, 85, 186-193.

Cross, S. E., & Madson, L. (1997). Models of the self: Self-construals and gender. Psychological Bulletin, 122, 5-37.

Darley, J., & Fazio, R. (1980). Expectancy confirmation processes aris- ing in the social interaction sequence. American Psychologist, 35, 867-881.

Derogotis, L. (1975). Affects Balance Scale. Towson, MD: Clinical Psychometrics Research Unit.

Downey, G. (1997, May). Rejection sensitivity and intimate relationships. Paper presented at the annual meeting of the American Psycho- logical Society, Washington, DC.

Downey, G., Bonica, C., & Rincon, C. (in press). Rejection sensitivity and conflict in adolescent romantic relationships. In W. Furman, B. Brown, & C. Feiring (Eds.), Adolescent romantic relationships. New York: Cambridge University Press.

Downey, G., & Feldman, S. (1996). Implications of rejection sensitivity for intimate relationships. Journal of Personality and Social Psychol- ogy, 70, 1327-1343.

Downey, G., Feldman, S., & Ayduk, O. (in press). Rejection sensitivity and male violence in romantic relationships. Personal Relationships.

Downey, G., Khouri, H., & Feldman, S. (1997). Early interpersonal trauma and adult adjustment: The mediational role of rejection sensitivity. In D. Cicchetti & S. Toth (Eds.), Rochester Symposium on Developmental Psychopathology: Vol. VIII. The effects of trauma on the developmental process (pp. 85-114). Rochester, NY: University of Rochester Press.

Downey, G., Lebolt, A., Rincon, C., & Freitas, A. L. (1998). Rejection sensitivity and children's interpersonal difficulties. Child Develop- ment, 69, 1072-1089.

Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to personality and motivation. Psychological Review, 95, 256-273.

Erikson, E. (1950). Childhood and society. New York: Norton. Eskin, M., & Kravitz, M. (1980). Child abuse and neglect. Washington,

DC: U.S. Department of Justice. Feldman, S., & Downey, G. (1994). Rejection sensitivity as a mediator

of the impact of childhood exposure to family violence on adult attachment behavior. Development and Psychopathology, 6, 231-247.

Gottman, J.M. (1979). Marital interaction: Experimental investigations. New York: Academic Press.

Gottman, J.M. (1993). The roles of conflict engagement, escalation, and avoidance in marital interaction: A longitudinal view of five types of couples. Journal of Consulting and Clinical Psychology, 61, 6 - 15.

Harris, M. B. (1993). How provoking! What makes men and women angry. Aggressive Behavior, 19, 199-213.

Hazan, C., & Shaver, P. (1994). Attachment as an organizational framework for the study of close relationships. Psychological Inquiry, 5, 1-22.

Heider, E (1958). The psychology of interpersonal relations. New York: Wiley.


Higgins, E. T., & Bargh, J. A. (1987). Social cognition and social perception. Annual Review of Psychology, 38, 369-425.

Hilton, J., & Darley, J. ( 1991 ). The effects of interaction goals on person perception. In M. N. Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 235-267). San Diego, CA: Academic Press.

Homey, K. (1937). The neurotic personality of our time. New York: Norton.

Jack, D. C. (1991). Silencing the self" Women and depression. Cam- bridge, MA: Harvard University Press.

Jack, D. C., & Dill, D. (1992). The Silencing the Self Scale: Schemas of intimacy associated with depression in women. Psychology of Women Quarterly, 16, 97-106.

Jussim, L. (1986). Self-fulfilling prophecies: A theoretical and integ- rative review. Psychological Review, 93, 429-445.

Jussim, L. (1991). Social perception and social reality: A reflection- construction model. Psychological Review, 98, 54-73.

Jussim, L., & Eccles, J. ( 1995 ). Naturally occurring interpersonal expectancies. In N. Eisenberg (Ed.), Social development: Review of personality and social psychology (Vol. 15, pp. 74-108). Thousand Oaks, CA: Sage.

Juvonen, J. (1991). Deviance, perceived responsibility, and negative peer reactions. Developmental Psychology, 27, 672-681.

Kayser, K. ( 1993 ). When love dies: The process of marital disaffection. New York: Guilford Press.

Kelley, H. (1979). Personal relationships: Their structure and processes. Hillsdale, NJ: Erlbaum.

Kenny, D., Kashy, D., & Bolger, N. (1998). Data analysis in social psychology. In D. Gilbert, S. Fiske, & G. Linzey (Eds.), Handbook of social psychology (4th ed., pp. 233-265). New York: McGraw- Hill.

Kobak, R., & Hazan, C. (1991). Attachment in marriage: Effects of security and accuracy of working models. Journal of Personality and Social Psychology, 60, 861-869.

Madon, S., Jussim, L., & Eccles, J. (1997). In search of the powerful self-fulfilling prophecy. Journal of Personality and Social Psychology, 72, 791-809.

McNulty, S.E., &Swann, W.E., Jr. (1994). Identity negotiation in roommate relationships: The self as architect and consequence of social reality. Journal of Personality and Social Psychology, 67, 1012-1023.

Merton, R. K. (1948). The self-fulfilling prophecy. Antioch Review, 8, 193-210.

Miller, D.T., & Turnbull, W. (1986). Expectancies and interpersonal processes. Annual Review of Psychology, 37, 233-256.

Miller, G. E., & Bradbury, T. N. (1995). Refining the association between attributions and behavior in marital interaction. Journal of Family Psychology, 9, 196-208.

Mischel, W. (1973). Toward a cognitive social learning reconceptualiza- tion of personality. Psychological Review, 80, 252-283.

Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality: Reconceptualizing situations, dispositions, dynamics, and invariance in personality structures. Psychological Review, 102, 246- 268.

Murray, S. L., Holmes, J. G., & Griffin, D. W. (1996a). The benefits of positive illusions: Idealization and the construction of satisfaction in

close relationships. Journal of Personality and Social Psychology, 70, 79-98.

Murray, S. L., Holmes, J. G., & Griffin, D. W. (1996b). The self-fulfilling nature of positive illusions in romantic relationships: Love is not blind, but prescient. Journal of Personality and Social Psychology, 71, 1155-1180.

Myers, J.L. (1966). Fundamentals of experimental design. Boston: Allyn & Bacon.

Rabiner, D., & Coie, J. (1989). Effects of expectancy inductions on rejected children's acceptance by unfamiliar peers. Developmental Psychology, 25, 450-457.

Reis, H. T., & Patrick, B. C. (1996). Attachment and intimacy: Compo- nent processes. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles ( pp. 523-563 ). New York: Guilford Press.

SAS Institute. (1989). SAS/STAT user's guide (Version 6, 4th ed., Vol. 2). Cary, NC: Author.

Simpson, J., Rholes, W. S., & Nelligan, J. S. (1992). Support seeking and support giving within couples in an anxiety-provoking situation: The role of attachment styles. Journal of Personality and Social Psy- chology, 62, 434-446.

Simpson, J., Rholes, W.S., & Phillips, D. (1996). Conflict in close relationships: An attachment perspective. Journal of Personality and Social Psychology, 71, 899-914.

Snyder, M. (1992). Motivational foundations of behavioral confirmation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 25, pp. 67-114). San Diego, CA: Academic Press.

Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Mar- riage and the Family, 38, 15-28.

Sroufe, L. A. (1990). An organizational perspective on the self. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood. The John D. and Catherine T. MacArthur Foundation Series on Mental Health and Development (pp. 281-307). Chicago: University of Chicago Press.

Stephen, T. D. (1984). Symbolic interdependence and post-breakup dis- tress: A reformulation of the attachment construct. Journal of Divorce, 8, 1-16.

Sullivan, H. S. (1953). The interpersonal theory of psychiatry. New York: Norton.

Swann, W. B., Jr. (1983). Self-verification: Bringing social reality into harmony with the self. In J. Suls & A. G. Greenwald (Eds.), Psycho- logical perspectives on the self (Vol. 2, pp. 33-66). Hillsdale, NJ: Erlbaum.

Swann, W. B., Jr. (1997). The trouble with change: Self-verification and the self. Psychological Science, 8, 177-180.

Veroff, J., Hatchett, S., & Douvan, E. (1992). Consequences of partici- paring in a longitudinal study of marriage. Public Opinion Quarterly, 56, 315-327.

Weiss, R. L., & Summers, K. (1983). Marital Interaction Coding System III. In E. Filsinger (Ed.), Marriage and family assessment (pp. 35- 115). Beverly Hills, CA: Sage.

Received July 18, 1997 Revision received February 17, 1998

Accepted March 4, 1998 •

The Self-Fulfilling Prophecy in Close Relationships: Rejection

Documents

Transcript of The Self-Fulfilling Prophecy in Close Relationships: Rejection